PRIMCED Discussion Paper Series, No. 60

Spatial Characteristics of Long-term Changes in

Indian Agricultural Production:

District-Level Analysis, 1965-2007

Takashi Kurosaki and Kazuya Wada

January 2015

Research Project PRIMCED Institute of Economic Research

Hitotsubashi University 2-1 Naka, Kunitatchi Tokyo, 186-8601 Japan

http://www.ier.hit-u.ac.jp/primced/e-index.html

Spatial Characteristics of Long-term Changes in Indian Agricultural Production: District-Level Analysis, 1965-2007

January 2015

Takashi Kurosaki† and Kazuya Wada‡

Abstract: In this paper, we comprehensively describe spatial patterns of long-term

changes in Indian agriculture at the district level. Variables of concern include the land

use intensity, the ratio of rice and wheat in areas under foodgrains, the ratio of

non-foodgrains in gross cultivated area, the fertilizer use intensity, and individual crop

shares in gross cultivated areas. As a byproduct of the descriptive analysis, we propose a

new regional classification of Indian districts based on their similarity in rainfall, the

initial cropping and land-use patterns, and the initial condition and changes in irrigation.

The proposed classification has a reasonable explanatory power in describing the spatial

patterns of long-term changes at the district level.

1. Introduction

Sustaining agricultural growth is key to rural development and poverty reduction in India. As

the room for extensive expansion has almost disappeared in Indian agriculture today, it is critically

important to improve land productivity to sustain the growth. Among various factors that contribute to

productivity improvement, the introduction of new technology has been investigated most intensively

in literature. For example, in the standard literature on long-term growth in agricultural production in

India, the contribution of Green Revolution since the late 1960s has been emphasized (e.g., Bhalla and

Tyagi 1989, Bhalla and Singh 2001, Bhalla and Singh 2009, Bhalla and Singh 2012). The Green

Revolution technology is characterized by high-yielding seeds, chemical fertilizer, and irrigation.

Another area on which the existing literature has focused is institutional aspects including land tenancy,

labor market institutions, and credit markets, as shown in articles published in previous issues of We thank V.K. Ramachandran, Yoshifumi Usami, Haruka Yanagisawa, and participants at the 2015 FAS conference and TINDAS workshops for their useful comments on earlier versions of this paper. We are also grateful to M.C.S. Bantilan, P. Parthasarathy Rao, and E. Jagadeesh for providing us with the DLS data. † Corresponding author. Institute of Economic Research, Hitotsubashi University, Tokyo, Japan. Phone: 81-42-580-8363, Fax: 81-42-580-8333, e-mail: kurosaki@ier.hit-u.ac.jp ‡ Faculty of Global Communication, University of Nagasaki, Nagayo-cho, Japan. Phone: 81-95-813-5739, Fax: 81-95-813-5739, e-mail: kwada@sun.ac.jp

1

Review of Agrarian Studies.

Besides these, there is another source of agricultural productivity growth, which is less

investigated in the literature. Even with little improvement in per-acre yield of individual crops, the

land productivity can increase through the reallocation of crops from low value-added to high

value-added crops and from regions where productivity is low to regions where productivity is high.

Using a longer-term horizon than adopted in the traditional literature, Kurosaki (2002, 2011, 2015)

shows that sustained growth in agricultural production began in India during the 1950s, much earlier

than the onset of Green Revolution, and shifts from low to high value crops (changes in cropping

patterns) contributed to the agricultural growth during the earlier growth period. Similar findings were

obtained for areas currently in Pakistan Punjab (Kurosaki 2003). Kurosaki (2003) also demonstrates

that crop shifts from low productivity districts to high productivity districts contributed to the

agricultural growth in West Punjab, especially during the colonial period. Nevertheless, there is a

dearth of empirical studies on the contribution of spatial crop shifts to productivity improvement in

post-independence Indian agriculture.

As a starting point for such studies, this paper describes spatial patterns of long-term changes

in Indian agriculture at the district level for a period from 1965 to 2007. The analysis employs a

district as the unit of investigation, defined by district boundaries prevalent in 1965. The purpose of

the paper is descriptive in nature, without rigorously investigating the role of technology or policies or

agrarian structure. Which districts produced which crops? How did such spatial patterns change over

time? We address these questions in this paper by combining various quantitative methodologies to

describe spatial changes.

A salient point of this paper is that such descriptive information is useful for addressing more

fundamental questions such as what kind of market and technology development characterizes Indian

agriculture. To understand the salience, microeconomic theory of spatial equilibrium (Takayama and

Judge 1971) is useful. Agricultural production is linked with consumption demand in general. This

linkage implies that when agricultural output markets are underdeveloped, farmers in a village produce

2

what villagers want to consume. This is a situation where spatial equilibrium is closed within a village

as a unit. The equilibrium is characterized by village-specific shadow prices,1 which may diverge

from market prices. Without technical innovation of producing individual crops, there is no room for

productivity improvement under this situation. As agricultural output and factor markets develop,

however, farmers and villages become more able to respond to the demand outside the village. By

shifting to crops whose value-added is higher if calculated using market prices, production value can

be improved even without innovation in individual crop production technology. If such market

development is accompanied by irrigation development, the room for individual farmers to respond to

market incentive becomes larger. The spatial pattern of agricultural production changes over time

reflects such market and technology development (Takayama and Judge 1971, Timmer 1997).

With this theoretical background, Kurosaki (2003) provides district-level analysis for the

case of West Punjab agriculture (now in Pakistan) for the period from 1903 to 1992. This paper shares

research motivation with Kurosaki (2003) but extends the analysis to the whole of India. As all-India

district-level analysis of agricultural production, Bhalla and Singh (2001) and Bhalla and Singh (2012)

are notable research outcomes. These studies, however, do not interpret the observed spatial changes

in the microeconomic framework of market development and spatial equilibrium. This focus

distinguishes this paper from these studies.

The rest of the paper is organized as follows. Section 2 explains the data used and shows the

heterogeneity observed across districts in terms of agricultural intensification. The heterogeneity

evidence is the first descriptive exercise on spatial characteristics of changes in Indian agriculture. As

the second descriptive exercise, Section 3 shows district-level GIS maps, which enables us eyeball

perusal of changes in spatial production patterns that occurred between 1965 and 2007. In Section 4,

we propose new agricultural zones derived from cluster analysis using the district-level data, which is

another way to aggregate spatial changes in descriptive analysis. Section 5 adopts a more parametric

1 See de Janvry et al. (1991) for how shadow prices are defined in mathematical models of farmers facing underdeveloped markets.

3

approach of describing the spatial changes, i.e., a regression analysis applied to district-level panel

data. The regression analysis identifies correlates of changes in intensification measures. Section 6

concludes the paper.

2. Data

2.1 Dataset Used

We use the district-level study (DLS) database compiled by the International Crops Research

Institute for the Semi-Arid Tropics (ICRISAT). The original data sources include government statistics

such as Agricultural Statistics of India and related publications at the state level. The compilation

procedure is reported in the DLS manual (ICRISAT 1998). Although our dataset is based on the

revised version up to 2007, the DLS manual has not been revised. The period of analysis is 42 years:

from agricultural year2 1965/66 to agricultural year 2006/07. Smaller districts where agricultural

production is negligible and statistics are reported only sporadically have been dropped from the

analysis. Several observations with inconsistent data have also been dropped.

As a result, we employ a balanced panel dataset of 311 districts spread over 19 major states

of India (Andhra Pradesh, Assam, Bihar, Chhattisgarh, Gujarat, Haryana, Himachal Pradesh,

Jharkhand, Karnataka, Kerala, Madhya Pradesh, Maharashtra, Orissa, Punjab, Rajasthan, Tamil Nadu,

Uttar Pradesh, Uttarakhand, and West Bengal) over the 42 years. The 311 districts were based on

district borders in 1965. According to the district borders in 2007, the 311 districts correspond to 498

districts. Regarding the state coverage, the analysis excludes 9 small states such as Jammu & Kashmir,

Sikkim, Goa, etc. and 7 federal territories, which are all minor in analyzing Indian agriculture. Figure

1 shows the spatial coverage of these 311 districts.

<Insert Figure 1 around here>

From the DLS database, we compiled the following variables for the analysis in this paper.

As production factors, we employ gross cultivated area (gca), net cultivated area (nca), irrigation ratio

2 An agricultural year of India refers to a period from July 1 to June 30.

4

(“net cultivated area, irrigated” divided by nca), quantity of fertilizer (the sum of N, P, K fertilizers),

the number of agricultural markets, the length of paved roads, and rainfall indicators. For individual

crops, we employ area and output quantity of rice, wheat, maize, sorghum (jowar), pearl millet (bajra),

finger millet (ragi), barley, chickpea (gram), pigeonpea (toor/arhar), and other pulses. The sum of

areas under these crops cover 60 to 70% of the gross cultivate area.3

As demonstrated by Kurosaki (2011) and Kurosaki (2015), the twentieth century Indian

agriculture can be characterized by sustained growth through improving land productivity and shifts to

higher value-added crops. These papers have shown that the index of land use intensity (=gca/nca), the

share of rice and wheat in the areas under foodgrain crops (srw), and the share of non-foodgrain crops

in the gross cultivated area (snfg) gradually increased throughout the century.

2.2 Spatial Heterogeneity in Agricultural Intensification

Has the increase in these measures in agricultural intensification occurred homogeneously in

all districts in India? As the first descriptive analysis of spatial characteristics, we plot in a histogram

the district-level trends of gca, intensity, snfg, and srw (Figure 2).

<Insert Figure 2 around here>

Figure 2 clearly shows a substantial inter-district heterogeneity. Although the four indices

were associated with positive trends at the all India level, the trend was negative for a non-negligible

number of districts. The heterogeneity is more substantial for snfg and srw than for gca and intensity.

This suggests that throughout India, gross cultivated area increased, mostly through the rising intensity

of land use, while the list of crops that occupied the increased area under cultivation differs from

districts to districts. Furthermore, the heterogeneity of trends in snfg and srw has been increasing in

recent years.4 In the next section, we examine which crops specifically were responsible for such

heterogeneity. 3 The dataset includes crop information for oilseeds, sugarcane, cotton, potato, onion, and fodder crops. In this paper, we aggregate them as non-foodgrain crops. Crop-wise analysis of non-foodgrain crops is left for further analysis. 4 Figure 2 was redrawn using the subsample of the period up to 1995. The re-drawn figure shows more compact distribution for snfg and srw. The redrawn figure is available on request from the authors.

5

3. Spatial Changes in Agricultural Production Described through GIS Maps

In this section, we describe spatial changes in agricultural production using GIS maps at the

district level. In other words, this is an eyeball investigation of spatial patterns. Each of Appendix

Maps shows four figures for each variable of interest. The upper left figure plots the initial distribution

in five quantiles, where the initial period refers to the three year average from 1965/66 to 1967/68. The

upper right figure (terminal, A) plots the terminal distribution in five quantiles, whose quantile

thresholds are the same as those used for the initial quantiles. The terminal period refers to the three

year average from 2004/06 to 2006/07. The lower left figure (terminal, B) plots the terminal

distribution in five quantiles, whose quantiles are re-defined over the terminal values. The comparison

of the initial and terminal A tells us about absolute changes in spatial patterns while the comparison of

the initial and terminal B tells us about relative changes in spatial patterns. The lower right figure plots

the distribution of growth rate. The growth rate was estimated for each district from 42 year data using

OLS. We include in the appendix maps that show interesting spatial changes. Maps not shown in the

appendix are available on request from the authors.

3.1 Cropped Area

Gross cultivate area (gca) increased in most of districts in India (Appendix Map 1). The

positive trend has been more significant in northern districts such as those in Rajasthan, Punjab,

Haryana, Madhya Pradesh, West Bengal, and Assam. On the other hand, several districts were

associated with negative trends in gca, many of which are found in Himachal Pradesh, Bihar, and

Tamil Nadu.

Looking at individual crops, rice-dominating districts in the initial years mostly remain as the

same in the terminal years (Appendix Map 2). Districts with higher trends in rice area than other

districts are concentrated in Punjab, Haryana, Chhattisgarh, and West Bengal. Except for those

districts in West Bengal, all others are located inland. On the other hand, districts with negative trends

6

in rice area are found in Tamil Nadu, Bihar, Jharkhand, and several districts in western India.

Wheat production is concentrated in north India (Appendix Map 3), spanning from districts

in Punjab and Haryana (henceforth, called “Punjab region”) to those in Uttar Pradesh, Bihar, Rajasthan,

Madhya Pradesh, and Maharashtra. Districts with higher trends than the national average are also

concentrated in the same states, except for Maharashtra. In Maharashtra, districts once cropped with

large area with wheat were associated with negative trends in wheat area, as in districts in south India

where wheat was not cultivated much in its traditional farming system.

A distinct spatial contrast is observed among coarse millets. In case of maize (Appendix Map

4), the initial production was concentrated in northern districts in Punjab, Haryana, Uttar Pradesh,

Bihar, Jharkhand, and Rajasthan. In most of these districts, maize area decreased since the mid 1960s.

New maize-producing centers have been emerging from interior Maharashtra, Madhya Pradesh,

Karnataka, interior Andhra Pradesh, and interior Tamil Nadu. The production of sorghum (jowar)

decreased in most districts throughout India, with no significant change in production centers in

districts south of Maharashtra (Appendix Map 5). The production of pearl millet (bajra) also

decreased in the majority of districts in India (Appendix Map 6). It is noteworthy of finding several

exceptional districts in eastern Rajasthan where bajra area has been increasing. The overall decline is

observed for ragi area as well, including those districts where ragi was once one of the most important

crops, such as those districts in Tamil Nadu, Karnataka, Andhra Pradesh, and Orissa (Appendix Map

7). In contrast to the overall declining trend, districts in Uttarakhand show an increase in ragi area.

Barley has becoming a minor crop in most of the districts, including those in Uttar Pradesh, where

barley was once occupied a substantial share of area under crops.

A significant spatial shift of production center has been observed for chickpea (Appendix

Map 8). The traditional production center in northern districts in Punjab, Haryana, Uttar Pradesh, and

Bihar has witnessed a rapid decline of area under chickpea. To replace these districts, new chickpea

production districts are appearing in Madhya Pradesh, Maharashtra, and northern Karnataka. In other

words, the production center for chickpea has been going south. Regarding pigeonpea, traditional

7

producing districts in Uttar Pradesh, Maharashtra, and Karnataka experienced a slight decrease in

areas under the crop. No new center of significance is emerging, however. Production of other pulses

is on the decline on average, except for several districts in Orissa.

3.2 Intensity of Land Use

As summary measures of cropping patterns focusing on the change in land use intensity,

Appendix Maps 9-11 plot intensity, srw, and snfg, which were already discussed in Figure 2 regarding

the heterogeneity among districts. By looking at maps, we can pin point where each of these measures

increased or decreased.

The variable intensity was high in initial years (1960s) in districts located in Punjab, Uttar

Pradesh, Orissa, and West Bengal (Appendix Map 9). This regional contrast remains the same in

terminal years during the 2000s. Trends in intensity are positive in the majority of districts, especially

in those districts whose initial level of intensity was high. Districts associated with a decline in

intensity are concentrated in Tamil Nadu and Himachal Pradesh.

The importance of rice and wheat in foodgrain production (srw) was high in initial years in

districts located in Punjab region, Uttar Pradesh, West Bengal, western Orissa, and coastal districts on

the Arabian Sea (Appendix Map 10). The trends in srw were highly positive in Punjab region and

Uttar Pradesh, while those were negative in Orissa. Changes in cropping patterns in favor of Green

Revolution crops occurred more in Punjab region. Furthermore, many of arid districts in Rajasthan,

where rice or wheat were not cultivated due to the lack of water, experienced a rapid increase in srw,

thanks to recent irrigation development. In northern and western parts of India, irrigation in arid and

semi-arid environments clearly favored these Green Revolution crops.

The tendency to grow pure cash crops is captured by variable snfg (Appendix Map11). In

initial years, snfg was high in western half of India, coming down from Punjab in the north to Tamil

Nadu and Kerala in the south. As trends in snfg were lower in Punjab while higher in western and

southern parts of India, snfg became higher in western and southern parts of India but no longer so in

8

northwestern India in the terminal years. In other words, the spatial change in cropping patterns was

heterogeneous across regions, reflecting different comparative advantage of each district among crops.

In some districts, the direction of change was towards Green Revolution crops, while in others, the

direction was moving away from Green Revolution crops.

3.3 Interpretation

The descriptive analysis in this section implies the following. First, regarding the area under

crops, the Indian agriculture already reached the limit of extensive expansion during the late 19th -

early 20th century (Kurosaki 2015). To overcome the limit, the land use intensity increased especially

in north India where irrigation was developed. Spread of chemical fertilizer was another factor that

contributed to the intensification of land use, but the spread was also facilitated by irrigation. During

the initial years of our analysis, irrigation was more developed in north India and coastal districts on

the Bengal Bay including those in Tamil Nadu. Almost everywhere since then, water availability

improved. The districts that experienced higher growth in land intensity were overlapping with the

districts where irrigation was developed faster.

Throughout the period of our analysis, we observed shifts of crops such as more areas under

rice and wheat in Punjab region and new production centers of coarse millets in the interior India.

These changes suggest that production specialization has been going on in response to comparative

advantages associated with heterogeneous climatic conditions and irrigation development. As

Kurosaki (2003) suggested, rural infrastructure such as roads and markets could be responsible for

these changes as well. According to the same study, there were two different phases of development of

agricultural output markets: the first with local market integration linking nearby villages and cities,

which occurred during the colonial period, and the second with national market integration after the

independence in 1947. The spatial changes observed during the period of our study suggest that the

process is still continuing during the post-colonial period. The next section examines the possibility of

new typology as a descriptive tool, which reflects the long-term changes mentioned above.

9

4. New Typology of Indian Agriculture

4.1 Empirical Strategy: Cluster Analysis Using District-Level Data

In the previous section, spatial changes observed in GIS maps were discussed using state

names as the main indicator for regional variation within India. However, in many cases, within-state

heterogeneity is significant, while in other cases, some districts in a state showed patterns more similar

to districts in the neighboring state than to the other districts in the same state. Using zones of regional

typology is thus a convenient tool to aggregate spatial changes in descriptive analysis.

In the literature, several zones of regional typology regarding Indian agriculture have been

proposed: state boundaries, as used in Section 3; fifteen agro-climatic zones designated by the Indian

ministry of agriculture; agro-climatic regions from B1 to B8 of the ICRISAT, which were employed in

ICRISAT (1998); a more recent attempt at the ICRISAT (Rao et al. 2004), discussed below; the

twenty-one ecological & agrarian regions for the Indian Subcontinent prepared by Thorner (1996), etc.

Given rich panel information included in our dataset, we attempt in this section to construct a new

typology exploiting the district-level information. We examine the usefulness of the new typology in

two ways: its ability to show coherent patterns (Subsection 4.2) and its explanatory power in

parametric regressions (Section 5). Before the examination, we explain our methodology to group

districts.

We adopt a quantitative methodology called “cluster analysis.” It is a general term

corresponding to the task of grouping a set of objects (in our case, districts) in such a way that districts

in the same group (called a “cluster”) are more similar to each other than to those in other clusters in

terms of several observable characteristics. To solve the clustering task, various computer algorithms

have been proposed and we have chosen a one that is popularly used in applied economics.5 Similar

cluster analysis has been adopted for India as well, for example by Rao et al. (2004).

5 Specifically, we adopt a hierarchical clustering algorithm based on Ward method using similarity of Euclid and the same weight for observable variables (standardized) in calculating the similarities. See Everitt et al. (2001) for methodological details. Command cluster in the STATA 10 software was used to obtain clustering results.

10

Regarding the observable characteristics used for the classification, we use 15 variables that

correspond to the initial conditions and one trend variable. The 16 variables are those described in the

previous sections. They include the initial values (average of the first five years of the period of our

study) of rainfall (annual, June, and July-August), irrigation ratio, land use intensity (intensity

analyzed before), and shares of 10 crops in the gross cultivated area, and the annual trend of the

irrigation ratio (obtained from time series regression for each district). Our strategy is thus to employ

pre-determined variables of production choices and exogenous technology variables in order to

describe the current production structure. In the classification exercise, we do not pay any attention to

the geographic contiguity. The cluster analysis results may show zones with geographically compact

areas; if the zoning predicts a zone comprising several districts that are not contiguous, such

information could be useful so that we should report as they are. This is the approach we adopt.

Our approach is in sharp contrast to the one adopted by Rao et al. (2004), whose detail is

described in ICRISAT (1999). They derived a regional typology with 15 zones (or 18 zones as two of

the 15 zones are further divided into subzones) using cross-section DLS data of averages of three years

from 1997/98. Their list of observable characteristics includes the shares of 15 crops and 5 livestock

products in the agricultural gross output value. They also allowed for different classification

depending on rainfed and irrigated regions and adjusted zone boundaries so that each zone is

geographically contiguous. As their procedure classifies districts according to the production mix

prevailing during the late 1990s, it provides us with a useful insight on the production structure

corresponding to that specific period. However, the choice of variables is mechanical and does not

reflect microeconomic reasoning of initial factors, endogeneity of crop choices, and market structure.

If the interest is on describing the structure at a specific period, their approach is justifiable but the

zoning procedure is better applied periodically with regular revisions. In other words, the procedure is

not very useful in inferring underlying, fundamental factors that affect spatial patterns of the long-term

dynamic changes. Our choice of the 16 observable variables is the result of our attempt to overcome

these shortcomings.

11

The algorithm applied to our data clearly suggests a coarse typology with 5 zones and a

medium-level typology with 10 zones, as shown in Figure 3. Unfortunately, more detailed typology

with zones more than 10 resulted in unstable classifications, depending on specific algorithms.

Therefore, we mainly adopt the medium-level typology with 10 zones in this paper.

<Insert Figure 3 around here>

4.2 Characteristics of New Typology Zones

What spatial patterns does each zone in our proposed typology show? If each zone does not

show coherent patterns, our new typology is of little value. We thus prepare Table 1, showing

characteristics of each zone derived from the cluster analysis. It specifically reports the spatial

distribution of each zone and its average values of the sixteen variables on which our clustering was

based. We describe the characteristics for each of the five large zones (L1 – L5), with explanations for

medium-level zones (M1-M10) within each large zone.

<Insert Table 1 around here>

Zone L1 contains districts where rice cultivation dominated during the initial years.

Therefore, we call it “rice zone.” The rice zone (L1) is subdivided into M1 districts in Orissa, West

Bengal, Assam, etc., and M2 districts in Andhra Pradesh, Tamil Nadu, etc. The two sub-zones are

distinguished by the amount of rainfall and the extent of irrigation development: M1 is thus called

“high rainfall, rainfed rice zone” while M2 is called “low rainfall, irrigated rice zone.”

Zone L2 (= zone M3) spreads into districts in hilly and coastal areas in Kerala, Assam,

coastal Andhra Pradesh, and coastal Karnataka. Due to extremely high rainfall, rice crops dominated

during the initial years as in Zone L1. However, L2 is distinguished from other zones by the high

extent of producing non-foodgrain crops. For this reason, we call L2 “extreme rainfall, rainfed,

non-foodgrains zone.”

Zones L3-L5 are characterized by semi-arid agriculture, distinguished by the extent of land

use intensity, irrigation ratio, and traditional crops. Using the most traditional crops in these districts,

12

we call L3 “semi-arid, extensive, wheat-pulse zone,” L4 “semi-arid, intensive, maize zone,” and L5

“rainfed, extensive, millet zone.”

Zone L3 “semi-arid, extensive, wheat-pulse zone” is further subdivided into M4 (semi-arid,

extensive, wheat-chickpea zone) that spreads over Madhya Pradesh, Uttar Pradesh, Haryana, and

Rajasthan, and M5 (“semi-arid, extensive, pigeonpea-barley zone) that spreads over Uttar Pradesh and

Madhya Pradesh. The key crop characterizing M4 is chickpea (gram) while the one characterizing M5

is pigeonpea (toor/arhar).

Zone L4 “semi-arid, intensive, maize zone” is similarly subdivided into M6 (semi-arid,

intensive, maize-dominant zone) and M7 (irrigation-intensive, wheat-maize zone). In M6, the

importance of maize in the traditional cropping patterns was more distinct than in L4. M7 may be

alternatively called “Punjab-type zone.” It contains districts in Punjab, Haryana, and western Uttar

Pradesh, where Green Revolution first spread during the late 1960s.

Zone L5 contains three sub-zones differentiated by the most important coarse millet crop.

M8 districts are located in Maharashtra, Madhya Pradesh, Andhra Pradesh, and Rajasthan, where

sorghum (jowar) dominated among the coarse crops during the initial years of our analysis. M9

districts are only found in Karnataka, where finger millet (ragi) was the dominant coarse millet. M10

districts are located in Rajasthan, Gujarat, and Maharashtra, characterized by the importance of pearl

millet (bajra) among coarse millets.

<Insert Table 2 around here>

While Table 1 shows the initial characteristics of each zone, Table 2 summarizes the trends

experienced in districts located in each zone and the terminal characteristics during the 2000s. It is

worth noting that the high level of land use intensity among L4 districts was maintained and these

districts are currently characterized by intensive use of land today. Especially, districts in M7

(Punjab-type zone) witnessed a growth rate of land use intensity and fertilizer use higher than other

regions. Looking at the rice wheat ratio (srw) or the non-foodgrain ratio (snfg), the zone-wise

difference is not very substantial. An exception is the rapid increase of srw in M10 (bajra zone),

13

reflecting the replacement of bajra by wheat as irrigation is developed. In districts in M10, fertilizer

use increased much faster than in other zones.

As shown above, the cluster analysis using 16 variables suggested a new spatial typology of

Indian agriculture. Each zone (or sub-zone) derived by the cluster analysis was associated with its own

initial conditions and changes thereafter. Therefore, we conclude that the new typology has an ability

to show coherent patterns in district-level descriptive analysis. In the next section, we examine the

usefulness of the new typology from a different angle as well.

5. Correlates of District-Level Changes in Land and Fertilizer Use

5.1 Empirical Model

So far in this paper, we found that four indicators of agricultural production intensity show

different spatial dynamics across districts and zones. The four indicators are intensity (gross cultivated

area divided by net cultivated area), srw (area share of rice and wheat in the total area under foodgrain

crops), snfg (areas under non-foodgrain crops divided by gross cultivated area), and fertilizer (per-acre

use of chemical fertilizer, total of N, P, and K). In this section, we estimate a parametric regression

model to identify correlates of district-level changes in these variables. The objective of the regression

exercise is again descriptive. We would like to quantify which districts experienced faster or slower

growth in the four intensification measures. Candidates for the correlates include state boundaries,

new zones suggested in the previous section, and more structural variables. As a byproduct of the

regression analysis, we can evaluate how much explanatory power our new typology has in descriptive

and parametric regression exercises. The regression model we estimate is specified as:

yit = ai + (b0 + Zibk)t + uit, (1)

where yit is one of the four indicators in district i in year t, a and b are parameters to be estimated, Zi is

14

a vector of variables that shift trends, and uit is a zero-mean error.6

As each district is associated with different time-invariant characteristics such as weather,

geography, history, etc., the level impact of such heterogeneity is perfectly controlled by district fixed

effects, ai. After controlling for such heterogeneity, which factor explains the diversity in district-level

growth rate? This is the main motivation of estimating equation (1). In other words, parameters in b

are of primary interest of this section. Equation (1) can be estimated by a standard one-way fixed

effect panel method. As Zi in the interaction term has no variation across time, we use district-clustered

robust standard errors to evaluate the statistical significance of parameter b.

We attempt four variants with respect to the choice of Zi. First, when Zi is specified as an

empty set, parameter b0 identifies the Indian average growth rates of the four indicators (Model A).

Then we include 18 state dummies in Zi as Model B and 9 zone dummies as Model C. In Models B

and C, we use the state (zone) where the largest number of districts are located as the reference,

corresponding to parameter b0. Then parameter bk shows how much faster or slower growth state

(zone) k experienced relative to the reference state (zone). In Model D, we include normalized

variables of initial intensity measures and exogenous technology and infrastructure variables in Zi.

Then parameter b0 shows the average growth rate corresponding to a hypothetical district that had the

average values of all variables in Zi while parameter bk shows the marginal impact the variable has on

the growth rate.

5.2 Regression Results

Regression results are shown in Table 3. As shown in Panel A, which correspond to Model A,

all of the four indicators had a positive trend, statistically significant at the 1% level. The land use

6 Equation (1) has its dependent variable in levels, not in their logs, and is estimated by weighted least squares (WLS). This is because our motivation of the regression analysis is descriptive, i.e., to obtain conditional means of district-level variables (intensity, srw, snfg, and fertilizer) that are aggregated consistently to the national average. By applying WLS to level variables with proper weights (nca for intensity, the total foodgrain area for srw, gca for snfg, and gca for fertilizer), we can achieve this consistent aggregation. Furthermore, the three of the four dependent variables are already in ratios (multiplied by 100), so that the coefficient estimates on the time trend have an intuitive meaning of average annual changes in percentage points. Regarding the fourth variable, fertilizer, it may be a good idea to take logs. However, the results are very similar when we use logs (full results are available on request from the authors).

15

intensity increased by 0.54 percentage points a year, srw by 0.33 percentage points, snfg by 0.25

percentage points, and fertilizer by 2.66 kg/ha per year.

<Insert Table 3 around here>

Panel B, Table 3 shows the regression results when each state is allowed to have a different

growth rate. As the number of districts in Uttar Pradesh (UP) is the largest, we use UP as the reference

state. The null hypothesis of homogeneous growth rates across states is rejected at the 1% level, as

shown in the last row of Panel B. Land use intensity (intensity) grew faster than UP in districts in West

Bengal, Punjab, and Haryana, while it grew slower by more than 0.5 percentage points in districts in

Tamil Nadu, Bihar, Chhattisgarh, Uttarakhand, and Jharkhand. In Punjab and Haryana, where intensity

grew faster, districts also witnessed faster growth in fertilizer. The importance of non-foodgrain crops

(snfg) increased faster than UP in districts in Maharashtra, Rajasthan, West Bengal, and Andhra

Pradesh. Using estimates for b0 and bk in Panel B, we examine which state shows a dynamic change

that was most similar to the one found at the national level (Panel A). Interestingly, all of the 19 states

had one or more variables out of the four that was associated with statistically-significant difference

from the national average. In this sense, no state in India represents the Indian average. In the relative

sense, however, we found that the dynamic changes observed in Gujarat were the most similar to the

national pattern, followed by Karnataka and Maharashtra.

Panel C, Table 3 shows the regression result when each zone in Table 1 is allowed to have a

different growth rate. As the number of districts in M8 (rainfed, extensive, jowar zone) is the largest,

we use the jowar zone as the reference zone. Again the null hypothesis of homogeneous growth rates

across zones is rejected at the 1% level. The variable intensity grew at the fastest rate in districts

belonging to M7 (Punjab-type zone), followed by districts belonging to M4 (semi-arid, extensive,

wheat-chickpea zone). In contrast, density grew at significantly lower rates in M9 (ragi zone) and M10

(bajra zone). Similar contrast is found for srw, snfg, and fertilizer. Although using a much smaller

number of explanatory variables, Model C explains the variation in data as good as Model B does, as

shown in adjusted R2 reported in Table 3. Therefore, we judge the ten-zone typology shown in Figure

16

3 as fairly useful. This does not imply that there will be other typology that has higher R2 than model C

and a fewer number of zones. The point here is that our typology, which was derived using the

criterion of utilizing the information on initial conditions and trends in irrigation only, has reasonable

explanatory power in parametric models for spatial changes, when we compare it with other existing

typologies.7

Finally, Model D employs structural factors as shifters of heterogeneous growth rates. In

other words, this is an attempt to open the black box represented by state- or zone-specific growth

rates by borrowing insights from microeconomic theory explained in the introduction. As structural

shifters, we utilize information contained in 8 variables: intensity (initial8 value), irrigation ratio noted

as iratio (initial and trend), srw (initial), snfg (initial), fertilizer (initial), rainfall (42 year average),

road density (initial), and market density (initial and dummy for missing information).9 The null

hypothesis of homogeneous growth rates across zones is rejected at the 1% level. The regression

results (Panel D, Table 3) clearly show that both the initial level and trend of iratio are the most

important determinant of heterogeneous growth rates in intensity, srw, snfg, and fertilizer. The initial

level of road density is associated with a higher growth rate of fertilizer while the initial level of

market density is associated with a higher growth rate of intensity. Therefore, the disparity in

infrastructure development during the 1960s resulted in the disparity in agricultural intensification

since then. Furthermore, the spatial dynamics of srw and snfg, which show different aspects of

commercialization of agriculture, are diverse across districts, reflecting the difference in the initial

conditions of cropping patterns and in rainfall patterns. Many of the coefficients on these variables

have opposite signs between srw and snfg. The adjusted R2 for Model D is similar to the one for Model

C, implying that the ten-zone typology has an explanatory power as high as a structural model.

7 In this paper, we compare our new typology (Panel C) and state boundaries (Panel B). We also estimated a similar model using the 18-zone typology of Rao et al. (2004). Adjusted R2 was 0.850 (intensity), 0.957 (srw), 0.810 (snfg), and 0.847 (fertilizer) (full results are available from the authors on request). These numbers are comparable to those reported in Panel C, Table 3. In this comparison as well, our new typology shows reasonable explanatory power in describing the spatial patterns. 8 “Initial” values in this regression are the averages of the first five years of the panel data. 9 We use the per-acre density of principal and sub markets in a district. As this information was missing for many districts in early years in Orissa, Bihar, and West Bengal, we included a dummy for the data missing as a shifter of growth rates.

17

6. Conclusion

In this paper, we described spatial patterns of long-term changes in Indian agriculture at the

district level using a balanced panel dataset from 1965/66 to 2006/07 (42 years). Comprehensively

investigating the land use intensity, the ratio of rice and wheat in areas under foodgrains, the ratio of

non-foodgrains in gross cultivated area, the fertilizer use intensity, and individual crop shares in gross

cultivated areas, we found the following.

First, there existed huge heterogeneity across districts in the speed of agricultural

intensifications in the last 42 years. Second, the eyeball perusal of GIS maps identified a shift of rice

production into the interior districts of north India, shift of wheat production to the east in north India,

new appearance of maize production centers in the interior districts of the Deccan, southward shift of

chickpea production, etc. The spatial shift appeared consistent with comparative advantages of each

district. Third, we attempted aggregating districts into similar zones using cluster analysis based on

their similarity in rainfall, the initial cropping and land-use patterns, and the initial condition and

changes in irrigation. The proposed classification has reasonable explanatory power in describing the

spatial patterns of long-term changes at the district level. Fourth, we estimated a parametric regression

model to identify correlates that were associated with heterogeneous growth rates of the land use

intensity, the ratio of rice and wheat in foodgrains, the ratio of non-foodgrains in gross cultivated area,

and the fertilizer use intensity across districts. The results confirmed the critical role of irrigation,

market, and road development in facilitating intensification of agricultural production. The regression

results also clarified the different aspects of agricultural commercialization represented by the rice

wheat share and the foodgrain share, respectively. These findings have enriched our knowledge on

spatial aspects of agricultural development in India.

The analysis in this paper is descriptive and preliminary in nature, however. Quantifying the

contribution of spatial changes to aggregate productivity improvement is left for further study. More

fundamental determinants of infrastructure and market development need to be examined in the

historical, institutional, and spatial context, which is also left for further research. In the current paper,

18

infrastructure and market development, including the key input of irrigation, were regarded as

exogenous to farmers’ decision making. This is unsatisfactory considering the political economy

context in which the development occurs. Another area of future work is more disaggregated analysis

combining household and village level changes in cropping pattern with changes at the district or state

(zone) levels. It is possible that the same change at the district level is observed in two districts despite

within-district, inter-village changes are substantially different in the two districts. Such cases will

shed further light to our understanding of the interaction between market development and agricultural

production. Extending the analysis to include more recent years is also left for further research.

19

References

Bhalla, G.S. and G. Singh, 2001. Indian Agriculture: Four Decades of Development. New Delhi: Sage Publications.

Bhalla, G.S. and G. Singh, 2009. “Economic Liberalisation and Indian Agriculture: A Statewise Analysis.” Economic and Political Weekly 46(52): 34-44.

Bhalla, G.S. and G. Singh, 2012. Economic Liberalisation and Indian Agriculture: A District-Level

Study. New Delhi: Sage Publications. Bhalla, G.S. and D.S. Tyagi, 1989. Patterns in Indian Agricultural Development: A District Level

Study. New Delhi: Institute for Studies in Industrial Development.

de Janvry, A., M. Fafchamps, and E. Sadoulet, 1991. “Peasant Household Behavior with Missing Markets: Some Paradoxes Explained.” Economic Journal 101(409): 1400-17.

Everitt, B.S., S. Landau, and M. Leese, 2001. Cluster Analysis, 4th ed. London: Arnold. ICRISAT [International Crops Research Institute for the Semi-Arid Tropics], 1998. District Level

Database Documentation --- 13 States of India, Volume I: Documentation of Files: 1966-94

Database (1966 district boundaries). Pattancheru, AP, India: ICRISAT. -----, 1999. Typology Construction and Economic Policy Analysis for Sustainable Rainfed Agriculture.

Pattancheru, AP, India: ICRISAT. Kurosaki, T., 1999. “Agriculture in India and Pakistan, 1900-95: Productivity and Crop Mix.”

Economic and Political Weekly, 34(52) December 25: A160-A168. -----, 2002, “Agriculture in India and Pakistan, 1900-95: A Further Note.” Economic and Political

Weekly, 37(30) July 27: 3149-3152.. -----, 2003. “Specialization and Diversification in Agricultural Transformation: The Case of West

Punjab, 1903-1992.” American Journal of Agricultural Economics 85(2): 373-387. -----, 2011, “Compilation of Agricultural Production Data in Areas Currently in India, Pakistan, and

Bangladesh from 1901/02 to 2001/02.” G-COE discussion paper, No.169/ PRIMCED discussion paper, No.6, Hitotsubashi University, February 2011.

-----, 2015, “Long-term Agricultural Growth in India, Pakistan, and Bangladesh from 1901/2 to 2001/2.” International Journal of South Asian Studies, Volume 7: 61-86 (in printing).

Rao, P. Parthasarathy, P.S. Birthal, Dharmendra Kar, S.H.G. Wickramaratne, and H.R. Shrestha. 2004. Increasing Livestock Productivity in Mixed Crop-Livestock Systems in South Asia. Patancheru, AP, India: ICRISAT.

Takayama, A. and G. Judge, 1971. Spatial and Temporal Price and Allocation Models. Amsterdam: North Holland Publishing.

Thorner, D. (ed), 1996. Ecological and Agrarian Regions of South Asia circa 1930. Karachi: Oxford University Press.

Timmer, C.P., 1997. “Farmers and Markets: The Political Economy of New Paradigms.” American

Journal of Agricultural Economics 79(2): 621-627.

20

Figure 1. Spatial Distribution of the 311Districts Analyzed

Source: Drawn by the authors using the DLS database (the same for following tables and figures). Note: The shaded area within thin lines corresponds to a district (boundaries in 1965) included in this study. Bolder lines show state boundaries in 2014.

21

Figure 2. Distribution of Average Annual Growth Rates at the District Level, 1965/66 - 2006/07

Notes: We first regress a time series model for each of the 311 districts, using the natural logarithm of gca, intensity, snfg, or srw as the dependent variable and the annual trend as the explanatory variable (gca = gross cultivated area; intensity = gca / net cultivated area; snfg = the share of non-foodgrain crops in gca; srw = the share of rice and wheat in the areas under foodgrain crops). We then plot the distribution of the 311 parameter estimates in a histogram. To make histograms easy to compare, we trim the range between -.02 (annual average decline at 2%) and .04 (annual average increases at 4%). The number of outliers outside the range is 2 for gca, 0 for intensity, 22 for snfg, and 10 for srw.

22

Figure 3. New Typology Zones Derived from Cluster Analysis

23

Table 1. Charactersitics of New Typology Zones for Indian Agriculture Derived from Cluster Analysis

Annual JuneJuly-August rice wheat maizesorghu

m

pearl

millet

finger

milletbarley

chick

pea

pigeon

pea

other

pulses

L1 (=M1,2) 83 OR, WB, TN, AP,

BH, CG, AS, JK,

MP, UP, HP, MH

0.20 0.19 -0.11 0.14 0.30 -0.41 1.02 -0.47 -0.28 -0.42 -0.31 -0.04 -0.19 -0.43 -0.41 0.12 Rice zone

M1 53 OR, WB, CG, AS,

BH, JK, MP, MH

0.42 0.51 0.25 0.04 -0.26 -0.31 1.18 -0.47 -0.17 -0.53 -0.38 -0.11 -0.23 -0.36 -0.38 0.18 High rainfall, rainfed rice

zoneM2 30 AP, TN, BH, UP -0.18 -0.37 -0.73 0.30 1.28 -0.59 0.72 -0.48 -0.48 -0.22 -0.18 0.08 -0.12 -0.55 -0.46 -0.01 Low rainfall, irrigated

rice zoneL2 M3 29 KL, AS, MH, KN,

WB

2.43 2.22 2.14 0.02 -0.18 -0.96 0.77 -0.80 -0.38 -0.64 -0.51 0.32 -0.50 -0.69 -0.58 -0.59 Extreme rainfall,

rainfed, non-

foodgrains zone

L3 (=M4,5) 48 UP, MP, HY, RS -0.32 -0.49 0.03 -0.10 0.01 0.96 -0.43 0.86 -0.33 -0.11 0.11 -0.26 1.17 1.65 1.06 -0.20 Semi-arid, extensive,

wheat-pulse zoneM4 30 MP, UP, HY, RS -0.35 -0.53 0.06 -0.37 -0.33 1.07 -0.68 1.14 -0.51 0.03 0.29 -0.28 0.15 2.23 0.66 -0.26 Semi-arid, extensive,

wheat-chickpea zoneM5 18 UP, MP -0.28 -0.43 -0.03 0.36 0.57 0.78 0.00 0.38 -0.03 -0.36 -0.18 -0.23 2.86 0.67 1.73 -0.11 Semi-arid, extensive,

pigeonpea-barley zoneL4 (=M6,7) 61 UP, PJ, UK, RS,

HP, BH, GJ, HY

-0.33 -0.28 -0.14 1.11 0.54 0.11 -0.40 0.93 1.29 -0.61 -0.26 -0.24 0.14 0.10 -0.32 -0.53 Semi-arid, intensive,

maize zoneM6 27 HP, RS, BH, GJ,

UK

-0.19 -0.07 -0.11 1.34 -0.29 -0.56 -0.49 -0.01 1.91 -0.58 -0.42 -0.20 0.20 -0.29 -0.39 -0.56 Semi-arid, intensive,

maize-dominant zoneM7 34 UP, PJ, HY -0.44 -0.45 -0.17 0.92 1.19 0.64 -0.33 1.68 0.79 -0.63 -0.12 -0.28 0.10 0.42 -0.27 -0.51 Irrigation-intensive,

wheat-maize zone

L5 (=M8,9,

10)

90 MH, RS, MP, KN,

GJ, AP

-0.57 -0.44 -0.51 -0.83 -0.58 0.10 -0.68 -0.40 -0.31 1.07 0.56 0.24 -0.38 -0.33 0.22 0.55 Rainfed, extensive,

millet zoneM8 57 MH, MP, AP, RS,

KN, GJ

-0.42 -0.31 -0.34 -0.83 -0.62 0.17 -0.65 -0.30 -0.21 1.73 -0.14 -0.18 -0.40 -0.18 0.66 0.40 Rainfed, extensive,

jowar zone

M9 7 KN -0.50 -0.71 -1.20 -0.70 -0.11 -0.50 -0.31 -0.89 -0.52 -0.22 -0.39 5.35 -0.54 -0.66 -0.06 1.02 Rainfed, extensive, ragi

zone

M10 26 RS, GJ, MH -0.93 -0.64 -0.70 -0.85 -0.63 0.10 -0.85 -0.48 -0.49 -0.04 2.36 -0.23 -0.28 -0.56 -0.67 0.77 Rainfed, extensive, bajra

zone

This table reports the normalized cluster-wise average. Therefore, under the normal distribution, the threshold for the top 5% (bottom 5%) is +1.64 (-1.64), while the threshold for the top 10% (bottom 10%) is +1.28 (-

1.28).

States included* Name (preliminary)

Large

（5zones）

Medium

（10

zones）

Number

of

districts

Rainfall Area under individual crop in gross cultivated area, initial value

* Name of states: AP=Andhra Pradesh, AS=Assam, BH=Bihar, CG=Chhattisgarh, GJ=Gujarat, HP=Himachal Pradesh, HY=Haryana, JK=Jharkhand, KN=Karnataka, KL=Kerala, MP=Madhya Pradesh,

MH=Maharashtra, OR=Orissa, PJ=Punjab, RS=Rajasthan, TN=Tamil Nadu, UP=Uttar Pradesh, UK=Uttarakhand, WB=West Bengal.

Intensit

y,

initial

value

Irrigatio

n ratio,

initial

value

Irrigatio

n ratio,

trend

24

Table 2. Changes That Occurred in Each Typology Zone Regarding Intensity of Agricultural Production and Cropping Patterns

intensity srw snfg fertlizer rice wheat maize sorghumpearl

millet

finger

milletbarley chick pea

pigeon

pea

other

pulses

L1 Rice zone OR, WB, TN, AP,

BH, CG, AS, JK,

MP, UP, HP, MH

( , , ) (+, , ) ( , , ) ( , , ) (+, , +) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , )

M1 High rainfall, rainfed

rice zone

OR, WB, CG, AS,

BH, JK, MP, MH

( , , ) (+, , ) ( , , ) ( , , ) (+, , +) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , )

M2 Low rainfall, irrigated

rice zone

AP, TN, BH, UP( , - , ) ( , , ) ( , , ) (+, , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , + , ) ( , , ) ( , , )

L2 M3 Extreme rainfall,

rainfed, non-

foodgrains zone

KL, AS, MH, KN,

WB

( , , ) (+, , +) (+, , +) ( , , ) ( , , ) ( , , -) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , )

L3 Semi-arid, extensive,

wheat-pulse zone

UP, MP, HY, RS ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) (+, , +) ( , , ) ( , , ) ( , , ) ( , , ) (+, , ) (+, , +) (+, , ) ( , , )

M4 Semi-arid, extensive,

wheat-chickpea zone

MP, UP, HY, RS( , , ) ( , , ) ( , , ) ( , , ) ( , , ) (+, , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) (+, , +) ( , , ) ( , , )

M5 Semi-arid, extensive,

pigeonpea-barley zone

UP, MP ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , +) ( , , ) ( , , ) ( , , ) ( , , ) (+, , ) ( , , ) (+, , ) ( , , )

L4 Semi-arid, intensive,

maize zone

UP, PJ, UK, RS,

HP, BH, GJ, HY

(+, , +) ( , , ) ( , , ) ( , , ) ( , , ) (+, , +) (+, , +) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , )

M6 Semi-arid, intensive,

maize-dominant zone

HP, RS, BH, GJ,

UK

(+, - , ) ( , , ) (+, , ) ( , , ) ( , , ) ( , , ) (+, , +) ( , , ) ( , , ) ( , , ) ( , , +) ( , , ) ( , , ) ( , , )

M7 Irrigation-intensive,

wheat-maize zone

UP, PJ, HY (+, + , +) ( , , +) ( , , ) (+, , +) ( , + , ) (+, , +) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , )

L5 Rainfed, extensive,

millet zone

MH, RS, MP, KN,

GJ, AP

( , , ) (-, , -) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) (+, , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) ( , , )

M8 Rainfed, extensive,

jowar zone

MH, MP, AP, RS,

KN, GJ

( , , ) (-, , -) ( , , ) ( , , ) ( , , ) ( , , ) ( , , ) (+, , +) ( , , ) ( , , ) ( , , ) ( , , ) ( , , +) ( , , )

M9 Rainfed, extensive, ragi

zone

KN ( , - , -) (-, , -) ( , , ) (+, , ) ( , , ) (-, , -) ( , , ) ( , , ) ( , , ) (+, + , +) ( , , ) ( , , ) ( , , ) (+, , )

M10 Rainfed, extensive,

bajra zone

RS, GJ, MH (-, , ) (-, + , -) ( , , ) ( , + , ) (-, , -) ( , + , ) ( , , ) ( , , ) (+, , +) ( , , ) ( , + , ) ( , , ) ( , , ) ( , + , )

Note: # (x, y, z) shows the relative positions of each cluster in the Indian average, where x indicates the initial value, y the trend, and z the terminal value. If the relative position is high (low) at the 20% significance

level, + (-) is shown. A black implies statistical insignificance. For instance, (+, -, ) indicates that the cluster had its initial value higher than the Indian average but the trend was more negative than the Indian

average, resulting in insignificant difference in the terminal value.

Large

（5zones

）

Mediu

m

（10

zones

）

States included

Indices for intensity of agric. production# Area under individual crop in gross cultivated area#

Name (preliminary)

25

Table 3. Regression Results for District-Level Changes in Agricultural Production Intensity

Summary StatisticsNo. of observations# 12,620 12,936 12,680 12,680Mean (Std.Dev.) of the dep. var. 126.43 (21.95) 52.01 (31.21) 32.27 (18.45) 55.39 (56.14)

Regression Results coeff. std.err coeff. std.err coeff. std.err coeff. std.err

A. Homogeneous Trends across All DistrictsDistrict fixed effects to the interceptCommon trend: b 0 0.544 *** 0.034 0.328 *** 0.030 0.254 ** 0.027 2.764 *** 0.104

R2 0.829 0.952 0.799 0.810

R2 adjusted 0.825 0.951 0.794 0.806

B. State-Specific TrendsDistrict fixed effects to the interceptBase trend: b 0 (reference=UP) 0.646 *** 0.054 0.738 *** 0.048 0.208 *** 0.040 3.592 *** 0.207

State-specific deviation of trends: b k (k = state dummy)

Andhra Pradesh -0.336 *** 0.096 -0.391 *** 0.112 0.223 ** 0.087 1.028 * 0.522Assam 0.019 0.153 -0.731 *** 0.049 -0.044 0.049 -2.369 *** 0.296Bihar -0.648 *** 0.132 -0.316 *** 0.082 -0.504 *** 0.185 -0.403 0.327Chhattisgarh -0.620 *** 0.077 -0.435 *** 0.058 -0.328 *** 0.065 -1.935 *** 0.395Gujarat -0.402 *** 0.078 -0.362 *** 0.097 0.132 0.131 -1.003 *** 0.309Himachal Pradesh -0.322 *** 0.109 -0.590 *** 0.052 -0.546 *** 0.139 -2.281 *** 0.247Haryana 0.509 *** 0.190 0.534 *** 0.152 0.143 0.136 1.102 * 0.612Jharkhand -0.582 *** 0.088 -0.575 *** 0.071 0.142 ** 0.069 -1.934 *** 0.328Karnataka -0.210 ** 0.098 -0.678 *** 0.087 -0.070 0.096 -0.719 * 0.399Kerala -0.499 *** 0.187 -0.708 *** 0.060 0.415 *** 0.085 -1.843 *** 0.300Maharashtra -0.038 0.110 -0.663 *** 0.065 0.190 *** 0.064 -1.258 *** 0.258Madhya Pradesh 0.100 0.099 -0.546 *** 0.086 0.203 ** 0.098 -2.113 *** 0.241Orissa 0.106 0.093 -1.030 *** 0.069 -0.127 * 0.065 -2.318 *** 0.289Punjab 0.809 *** 0.142 0.088 0.112 -0.538 *** 0.066 1.020 *** 0.244Rajasthan -0.115 0.091 -0.474 *** 0.088 0.245 ** 0.095 -2.411 *** 0.287Tamil Nadu -0.779 *** 0.087 -0.723 *** 0.096 0.068 0.103 0.399 0.406Uttarakhand -0.626 *** 0.154 -0.818 *** 0.088 -1.596 *** 0.523 -0.733 1.850West Bengal 0.881 *** 0.173 -0.510 *** 0.074 0.225 *** 0.051 0.082 0.313

R2 0.871 0.969 0.827 0.873

R2 adjusted 0.867 0.968 0.822 0.870

F (18, 310) stat for b k =0 for all k . 20.08 *** 28.66 *** 17.47 *** 37.48 ***

C. Typology-Zone-Specific Trends (10 zones)District fixed effects to the interceptBase trend: b 0 (reference=8.Jowar

zone)

0.584 *** 0.057 0.188 *** 0.045 0.407 *** 0.059 2.526 *** 0.177

Zone-specific deviation of trends: b k (k = cluster dummy)

1. High rainfall, rainfed rice zone -0.003 0.121 -0.075 0.065 -0.300 *** 0.077 -0.372 0.2542. Low rainfall, irrigated rice zone -0.451 *** 0.103 0.087 0.084 -0.300 ** 0.118 1.972 *** 0.3183. Ext. rainfall, rainfed, non-fg zone -0.163 0.169 -0.099 0.063 0.047 0.078 -0.228 0.3294. Semi-arid, ext., wheat-chickpea zone 0.224 * 0.117 0.327 ** 0.156 0.049 0.086 -0.247 0.3145. Semi-arid, ext., pigeonpea-barley zone0.121 0.085 0.779 *** 0.075 -0.411 *** 0.079 0.523 0.3566. Semi-arid, intensive, maize zone -0.366 *** 0.093 0.068 0.095 -0.722 *** 0.124 -0.426 0.3377. Punjab type zone 0.373 *** 0.128 0.578 *** 0.065 -0.330 *** 0.092 2.037 *** 0.2769. Rainfed, extensive, ragi zone -0.357 *** 0.081 -0.114 0.094 0.142 0.088 0.650 0.49810. Rainfed, extensive, bajra zone -0.194 ** 0.086 0.009 0.070 0.039 0.089 -0.979 *** 0.307

R2 0.845 0.962 0.818 0.855

R2 adjusted 0.841 0.961 0.813 0.851

F (9, 310) stat for b k =0 for all k . 10.53 *** 31.06 *** 14.52 *** 20.08 ***

(Yes.$)(Yes.$) (Yes.$) (Yes.$)

intensity (x100) srw (x100) snfg (x100) fertilizer

(Yes.$) (Yes.$) (Yes.$) (Yes.$)

(Yes.$) (Yes.$) (Yes.$)(Yes.$)

26

Table 3. Regression Results for District-Level Changes in Agricultural Production Intensity (cont'd)

coeff. std.err coeff. std.err coeff. std.err coeff. std.err

D. Initial-Factor-Dependent TrendsDistrict fixed effects to the interceptBase trend: b 0 0.530 *** 0.048 0.339 *** 0.030 0.257 *** 0.032 2.658 *** 0.102

Deviation for the following variables: b k (k = structural factors)

intensity, initial (normalized) -0.109 0.072 0.113 *** 0.031 -0.207 *** 0.051 -0.017 0.091iratio, initial (normalized) 0.181 ** 0.078 0.251 *** 0.043 0.102 ** 0.046 0.485 *** 0.114iratio, trend (normalized) 0.224 *** 0.043 0.220 *** 0.025 0.101 *** 0.029 0.478 *** 0.073srw, initial (normalized) 0.085 0.056 -0.113 *** 0.036 -0.178 *** 0.038 -0.130 0.103snfg, initial (normalized) -0.005 0.048 -0.001 0.035 -0.053 0.039 -0.174 ** 0.084fertilizer, initial (normalized) -0.136 ** 0.052 -0.102 *** 0.035 0.019 0.030 0.946 *** 0.119rainfall, 42 year average (norm.) -0.030 0.065 -0.017 0.039 0.227 *** 0.038 -0.001 0.145road density, initial (norm.) -0.078 0.148 0.074 0.075 -0.076 0.075 0.714 ** 0.319market density, initial (norm.) 0.169 *** 0.063 0.041 0.033 -0.014 0.035 0.031 0.169dummy for missing market info 0.008 0.064 -0.079 * 0.042 -0.081 0.056 0.273 * 0.142

R2 0.844 0.966 0.818 0.877

R2 adjusted 0.840 0.965 0.813 0.874

F ( 10, 310) stat for b k =0 for all k . 6.43 *** 20.73 *** 6.64 *** 27.62 ***

# Although we potentially use 13,062 observations (=311 districts * 42 years), the actual number of observations used in

regressions was less than this due to missing observations.

(Yes.$)

fertilizerintensity (x100) srw (x100) snfg (x100)

(Yes.$) (Yes.$)

$ For brevity, district fixed effects on intercepts are not reported. They are jointly significant at the 0.1% level in all specifications.

Notes: Estimated by weighed least squares with district-clustered robust standard errors. The weights are nca for intensity , the

total foodgrain area for srw , gca for snfg , and gca for fertilizer ). Statistically significant at 1% ***, 5% **, 10% *. See Table 1 for

the distribution of districts across 10 zones used in Panel C. The summary statistics for original variables used to calculate shifters

in Panel D are available on request from the authors.

(Yes.$)

27

28

29

30

31

32

33

34

35

36

37

38